Perhaps the most fundamental task in dealing with computer software is deciding which software to use. Often momentum narrows the field but in other cases there may be a wide range of options without evident reasons to favor some over others.
One helpful means to discover projects that are likely to have active community engagement and wide support is to look at projects which have been accepted by larger initiatives. For example graduated projects from the Cloud Native Computing Foundation are likely to be strong contenders when working in (unsurpisingly) cloud native environments.
Million Dollar Lines of Code provides some decent war stories around lack of cost management and lands on the useful formula of tracking what cost efficiency rather than blindly seeking cost reduction. It seems somewhat appropriate as an introduction to the concerns but seems to leave some gaps as it focuses on fairly discrete examples and then jumps to a metric without providing strong guidance around more consistent principles and monitoring (there’s a related concept I need to extract that draws on things like Murphy’s Law, Safety II, and Release It!). The latter also seems notable in that the premature optimization quote is mentioned (attributed to Hoare) but without an actionable less “evil” alternative.
I’ve been in organizations with different motivations to pursue governance, and operating within an architecture role governance is often something which significantly informs proposed designs and for which I often find myself acting as a mediator. Governance like other hygiene-style concerns such as security must be integrated into core work rather than perceived as bolted on friction, and the overarching governance perspective should be complementarily agile to core processes rather than introducing an impedence mismatch.
(Among many other things) containers promise resource utilization optimization, but unfortunately many organizations think such benefits can be realized without proper attention and potentially making deeper discussions around tradeoffs. Among other issues I’ve witnessed organizations that didn’t understand the underlying concepts pursue aggressive bin packing which resulting in behaviors which not only negatively impacted performance but was almost certainly counter-productive due to effectively starving nodes and introducing spurious restart churn; in the most egregious case this was also compounded by steady double digit CPU steal.
“Operational Excellence” feels like a relatively new word for what should be standard practice, but is too often neglected. Contrary to how many companies seem to approach things, creating software is almost invariably solely a cost and the value is delivered at the time the software is used or run. Operating the resulting software is therefore the primary objective and development is in service of that, and notions such as excellence should fall out of correcting any related priority inversion in a healthy company that seeks a culture of quality.
A good laundy list of operational excellence concerns is provided here.